The best AI users I have watched are not the most technical people in the room. They are the best communicators. And once I saw that pattern clearly, a second pattern came into focus behind it: the way women tend to communicate is the way AI wants to be used.
I know where some readers went when they read that sentence. I am aware this is a controversial thing to write, and I am writing it anyway, because I keep seeing it and I cannot unsee it.
Stay with me. This is not an article about who is better. It is an article about what AI actually rewards, and why most organizations are looking for their AI talent in the wrong place.
An Off-Road Course Showed Me Something I Could Not Unsee
A few years ago I did the Land Rover Experience near my home in Carmel Valley. A full day of off-road driving on a purpose-built course: steep grades, side tilts, water crossings, obstacles engineered to teach you what the machine can really do. It was one of the most fun days I can remember.
The exercise I still think about came partway through the day.
They blindfolded the driver.
The passenger got out of the vehicle and stood outside with the windows down. Their job was to talk the driver through a course of cones. The driver could see nothing. The guide could not touch the wheel. Every successful maneuver had to travel through one channel: language.
Before we started, the instructor told us something interesting. This exercise always works better, he said, when the woman is the one outside giving the directions and the man is the one behind the wheel. Not sometimes. Always. His explanation was simple: women communicate better.
He was not making a political statement. He was reporting the data from running the same exercise, week after week, with hundreds of couples.
And when you watch the exercise, you can see why. The good guides did not bark commands. They built a shared picture. Small correction, pause, confirm, next correction. They noticed when the driver tensed up and slowed the pace of instruction. They checked understanding before the cone, not after it. They adjusted their language to the person receiving it, in real time, continuously.
The guide's skill was not driving. The guide's skill was making another mind see what she could see.
I have been thinking about that blindfolded Land Rover a lot lately, because I finally realized what it is.
It is the most accurate model of working with AI I have ever encountered.
The Interface Is Not Technical. The Interface Is You.
Think about what AI actually is from the user's seat. It is the blindfolded driver: enormous capability, zero view of your world. It cannot see your business, your constraints, your standards, or the outcome you are holding in your head. You are the voice outside the window. The quality of what the machine does is bounded by the quality of what you can convey.
AI is the first consequential technology in history whose interface is communication itself.
Every prior technology asked you to learn its language. Software made you learn menus and syntax. Databases made you learn queries. Even the smartphone made you learn its gestures. AI inverted that. It speaks yours. Which means the leverage point moved. The skill that separates a mediocre result from an exceptional one is no longer technical fluency. It is the ability to give context, state intent, read the response, and refine. Conversation, in other words. Iteration.
Now I want to be precise about what I mean by the feminine part of this argument, because this is where the controversy lives.
I am not making a claim about biology, and I am not saying men cannot do this. I am describing a communication pattern that anyone who has spent time in rooms full of people has watched with their own eyes. Women, broadly and observably, connect by communicating. A woman feels most connected in a conversation when she understands the person across from her and feels understood in return. Understanding and being understood is the loop she is naturally running, all the time, in every interaction that matters to her.
That loop has a name in AI work.
It is called iteration. It is the exact behavior that produces great results with these tools: offer context, check what came back, notice the gap between what you meant and what was heard, close the gap, go again. The people who run that loop instinctively do not need to be taught how to use AI. They need to be handed the keys.
For a century we built workplaces that rewarded a different pattern: command, decide, transmit, move. That pattern had its era, and the era is ending. There is more feminine energy entering the workplace, and that is not a cultural accident. The economics moved. The tools at the center of work now reward connection over command.
The People Built for This Tool Are the Least Likely to Use It
Here is the part of this story that keeps it from being a feel-good observation, and the reason I am writing it.
In May of this year, researchers from Harvard Business School, Stanford, and Berkeley published the largest synthesis of evidence we have on who is actually using generative AI. They pulled together 76 sources covering more than 100 countries and nearly 319,000 respondents. The finding: 47.8 percent of men use generative AI, versus 39.3 percent of women. A relative gap of roughly 22 percent, present in nearly every country measured. The gap has narrowed since 2023, then stabilized. It has stopped closing on its own.
Read that against everything above.
The group whose native communication pattern most closely matches what this technology rewards is the group using it least.
The researchers opened their paper by noting that AI should especially benefit people who have faced institutional barriers to building skills at work, women most of all. Then their data showed those same women adopting it at the lowest rates. And when they dug into why, the reasons were not about capability. Women report less confidence prompting, more worry that using AI will be seen as cutting corners, more concern about how it reflects on their competence.
The gap is not a skills gap. It is a permission gap.
I watch the other side of that gap in my consulting work. At an insurance agency I work with, the first person to produce something remarkable with AI was not a technical hire. She was a senior producer who had never touched the platform before that day. She spoke her quoting process out loud, the process she knew cold after years of doing it, and the AI turned her clarity into a near-perfect procedure document on the first pass. One factual correction and it was usable. Their operations manager put words to what the whole room was feeling: "It's clicking. I didn't know how good I would get it, but it's clicking."
Nobody in that room wrote a line of code. The skill on display was articulation. Knowing the work, and being able to say it.
Stop Running AI Adoption as a Technical Project
If you lead an organization, here is what I want you to do with this.
Look at how your AI rollout is structured right now. In most organizations I walk into, it is framed as a technical project. IT owns it. The pilot group is the technical team. The training covers features and tools. The unstated assumption underneath all of it: the people who are good with technology will be the people who are good with AI.
That assumption is wrong, and it is quietly costing you your best AI talent.
AI adoption is not a technical project. It is a communication skill being distributed across your workforce. Run it that way, and different names rise to the top of the pilot list:
- The person whose emails never get misread
- The person who can explain a messy process to a new hire without notes
- The person clients calm down for
- The person who asks the question in the meeting that untangles the whole discussion
In my experience those people are disproportionately women, and in most organizations they are nowhere near the AI pilot group. Meanwhile, the adoption data says many of them are quietly holding back, waiting for a signal that using these tools counts as competence rather than a shortcut.
You can send that signal this week. Name your actual criteria for AI champions out loud: we are selecting for communication, context, and judgment, not technical background. Then look at who qualifies.
If You Are a Man Reading This, the News Is Good
This is not a zero-sum observation, and it is not a verdict on anyone.
The most common failure mode I see with AI is men treating it the way they were trained to treat software: issue a command, take the first output, judge the tool. One shot, no context, no loop. The result is mediocre, the conclusion is "AI is overhyped," and the real lesson goes unlearned.
The lesson is that communication is a skill, skills are learnable, and this one now has a direct line to output quality. Be the voice outside the window. Give the machine the context you would give a smart colleague on their first day. Check what came back against what you meant. Close the gap and go again. Every man I have watched make that shift got dramatically better results within days, because the model was never the constraint. The conversation was.
The Land Rover instructor was not telling us who was smarter. He was telling us which skill the exercise measured.
AI is the same exercise, running at global scale, and it measures the same thing.
The driver had the wheel, the pedals, and four hundred horsepower. The one who got the vehicle through the cones was the one who knew how to talk.
